On advice

Most people are eager to ask for (good) advice. Many people make a living on giving advice - just look at how many self-help books are out there. But how do you know which advice to follow?


Here are some common criteria to tell good advice from bad ones. 


  1. Does this advice itself follow logical reasoning and “make sense”?

  2. Is there a “consensus” about this advice? 

  3. Does this advice come from a “successful” person?


The problem is there are good counter arguments against all the above. 


  1. Availability bias and belief bias. Think about a typical i/o system. The input is available information to a person, the model is the person’s belief system, the output is the conclusion whether this advice makes sense or not. Two people who believe they are both rational beings could easily arrive at opposite conclusions. 

  2. There’s an old saying “the truth lies in the hands of a few”. The appeal of good advice is that it’s not too obvious, contrarian or even controversial. If everyone agrees on this advice, then it becomes common sense.

  3. Halo effect. Too many examples of this. 


There’s also the problem with hindsight bias. If someone is successful in their field, their advice on how to be successful in this field mostly describes what they have done. The problem is this type of advice often discounts randomness. Again, using the i/o system, even the exact same input (effort) will most likely result in different outcomes, given the randomness of the model [1]. 


I think the only way to know if you should follow a piece of advice is to try for yourself. If your model works with the advice, then it is good advice for you. Some advice may be hard to try because it takes a lot of effort, then break it down to smaller components and try a single component at a time. The trick here is to ruthlessly disqualify advice as soon as they don’t work out, but not before the minimum time that is required to reveal progress. Say you want to try running for 30 minutes a day to lose weight, if you just run once or one day, then obviously you are not going to see any positive result. But if you run a week, or a month, then it gives you a good trial to see if it works for you. If not, don’t continue. So this really is  an estimation problem - how much time/resource should be assigned to try out a new advice? That’s the real question to tell if advice is good for you or not.


Like any other estimation problem, you get better at it from experience. Then there may be a pattern to be seen of the minimum resource required to test out the effectiveness of an advice. Being persistent on trying out different advice but quickly dropping a single ineffective advice is the key. It could even be a goal to try x pieces of advice each month and increase x from low to high gradually. 


Be curious.


[1] Here the model refers to a non-deterministic view of the universe.


Comments

Popular posts from this blog

Don’t start believing

It's just my opinion

Are you having fun